{"id":"W4385572319","doi":"10.18653/v1/2023.acl-industry.55","title":"Weighted Contrastive Learning With False Negative Control to Help Long-tailed Product Classification","year":2023,"lang":"en","type":"article","venue":"","topic":"Text and Document Classification Technologies","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"Queen's University","funders":"","keywords":"Categorical variable; Categorization; Computer science; Regularization (linguistics); Artificial intelligence; Task (project management); Taxonomy (biology); Machine learning; Pattern recognition (psychology); Natural language processing; Engineering","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007910512,0.002186514,0.002139214,0.001606493,0.0009738666,0.001920611,0.004101744,0.002963451,0.003193386],"category_scores_gemma":[0.02398996,0.0004405977,0.0009738596,0.001353138,0.002432975,0.004217053,0.002975808,0.004195451,0.001303458],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001776126,"about_ca_system_score_gemma":0.001647503,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002363401,"about_ca_topic_score_gemma":0.003011922,"domain_scores_codex":[0.9957074,0.001398923,0.0002744871,0.001155904,0.001086369,0.0003769874],"domain_scores_gemma":[0.9864997,0.007907549,0.001141154,0.001677159,0.002335639,0.0004387703],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00197977,0.001260642,0.01215957,0.0005204159,0.0002951886,0.000643692,0.0003417615,0.1770833,0.02417455,0.02108616,0.01982522,0.7406297],"study_design_scores_gemma":[0.00005837771,0.0002059152,0.0008421295,0.00002927587,0.00004537862,0.0001607482,0.00003148761,0.9817613,0.006702794,0.008611333,0.001530562,0.00002064448],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07178314,0.001708086,0.9188817,0.001099727,0.0003132631,0.0002327757,0.0002818207,0.0027009,0.002998566],"genre_scores_gemma":[0.7947029,0.0003972355,0.1947737,0.001391524,0.0003969647,0.000330446,0.001257091,0.0003308092,0.006419199],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007910512,"threshold_uncertainty_score":0.04183525,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01927232653048668,"score_gpt":0.2567327726888022,"score_spread":0.2374604461583155,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}